by is Mooney
On top of that, it’s also possible that global warming “deniers” reason in a more motivated way on this issue than the “accepters” do, although our data on this point are not as conclusive as they were on nuclear power. What’s clear is that after reading our essays—essays that either supported or opposed our subjects’ initial views about whether global warming is real and caused by humans—the two groups did indeed respond differently.
On our first measure of motivated reasoning—remember, this was the “spread” between how persuasive a friendly essay was and how unpersuasive an unfriendly essay was—those who denied climate science appeared to show a larger gap (thinking the global-warming-is-bunk essay was persuasive and the global-warming-is-real essay was unpersuasive) than the accepters did (thinking the global-warming-is-real essay was persuasive while the global-warming-is-bunk essay was unpersuasive).
This finding was not very strong or statistically significant for all participants in the study. But it became increasingly strong, and increasingly significant, as our subjects’ political knowledge increased. Thus for instance, global warming believers who answered one political knowledge question right were about 10 percent more likely to call a view different from their own persuasive (p = .03), and those who answered two political knowledge questions right were 15 percent more likely (p = .006). Thus, you might say that as global warming deniers’ level of political knowledge increased, so did their bias, leaving the more knowledgeable deniers considerably more motivated than the more knowledgeable believers on our first measure. And this result was statistically significant.
On our second measure of motivated reasoning—whether your opinion changed after reading an essay that challenged your preexisting beliefs—the result is more complicated. First, we found that those who accepted human-caused global warming were more resistant to the (bogus) essay we created trying to debunk it. However, this result was not statistically significant.
As political knowledge increased, however, deniers were just as resistant to changing their minds after reading warming-is-real essays as accepters were after encountering warming-is-a-hoax essays. (And note: This means the smarter deniers were getting more and more convinced about a factually wrong belief.) Here again, however, the finding did not reach statistical significance.
So on our first motivated reasoning item (but not our second), increasing political knowledge is significantly associated with increasing motivated reasoning for global-warming deniers—but not for global-warming accepters. And this is consistent with what Chris likes to call the “smart idiot” effect—conservatives who are more knowledgeable, or more politically engaged, becoming more biased.
Based on the results reported so far, our hypothesis was faring pretty well. Conservatives were less Open, and were considerably more biased political reasoners on nuclear science—the issue where liberals are supposedly more biased. Plus, science deniers were more motivated than science accepters on at least one measure of MR regarding global warming—and they were motivated in favor of being factually wrong.
However, when it came to non-political items, this pattern didn’t hold any longer.
In fact, we did not achieve a satisfactory measure of general, across-the-board motivated reasoning as a trait or individual tendency. There is no question that motivated reasoning was happening on almost every item. For most of the items, our subjects found friendly essays more persuasive than unfriendly essays. What’s more, upon closer analysis, most of the items showed that taking a strong view on the topic was associated with higher levels of motivated reasoning than taking a moderate view. In other words, there was a strong prior attitude effect, which is just what we would expect to find if motivated reasoning was going on.
However, on these non-political items—involving how good one’s favorite quarterback is, how good one’s school is, and so on—motivated reasoning seemed to be driven less by any trait that the individual possessed, and more simply by this prior attitude effect: A person’s emotional investment in a particular idea, belief, or “attitude object.” In other words, we didn’t find strong evidence that motivated reasoning even exists as a chronic personality trait that can be measured on a multi-item index or scale. If he’s emotionally attached to a thing, a person is probably going to be motivated in his reasoning about it, at least when initially confronted with “evidence” about it. (It may well be that more open-minded people will continue thinking about things after the initial shock, and may later weigh additional evidence, whereas more closed-minded people will regard the matter as closed—but we could not test that in a single sitting.)
In other words, this is where our study tried to go considerably farther than prior work—to shoot for the moon—and we didn’t succeed. We wanted to show that the same patterns of motivated reasoning that occurred on political issues would also hold up on a battery of non-political issues—indeed, that they would hold across the board. If so, this would mean that conservatives are simply more defensive about their prior beliefs, period.
At least in this first pass at the question, it didn’t work out that way.
But while we didn’t find conservatives systematically engaging in more motivated reasoning than liberals on non-political issues, there was something intriguing that we did find. And this was, by far, our strongest result of all.
We measured how much time our subjects spent reading the essays. We did so simply because Everett’s mentors in political science, Charles Taber and Milton Lodge of Stony Brook, had found that more time spent looking at counter-attitudinal arguments was associated with stronger motivated reasoning, presumably because people spent more time arguing in their heads against the contrary evidence.
While our study design was substantially different from theirs, we wondered if increased reading time would again be associated with increased motivated reasoning. It actually wasn’t, in our sample. But we anticipated there might be a complication with this idea: Liberals might spend more time reading the essays. And guess what: they did!
4. Conservatives spend less time attending to new information than liberals do. Across the twelve items in our study—both political and non-political—the tendency to spend more time on a particular page of essay formed a very reliable “scale,” regardless of whether we measured respondents’ reading of pro-attitudinal essays, counter-attitudinal essays, or all essays. This remained the case even after throwing out participants who spent so little time reading an essay that they could not have possibly attended to it at all—and horrors, sometimes students actually do this, clicking mindlessly through our surveys for extra credit without actually reading them.
We then tested whether conservatives spend less time reading the essays—and at quite robust levels, they do. For example, a strong self-identifying conservative was estimated to spend an average of 10 seconds less than a strong liberal looking at a single “screen page” of essay material!
Indeed, every single measure of conservatism we had was significantly correlated with less reading time, and in some cases highly significantly: self-identified general ideology (r = 0.30, p =.0003), self-identified social/moral conservatism (0.31, p =.001), self-identified fiscal conservatism (0.21, p =.03), issue-position-derived index of moral conservatism (0.21, p =.01), issue-position-derived index of fiscal conservatism (0.20, p =.016), issue-position-derived index of “toughness-issue” conservatism (0.19, p =.026), Republican party identification (.24, p =.004), and most powerfully of all, authoritarianism (0.32, p =.0001).
Let us unpack what that last finding, in particular, means. If this result—that authoritarians spend significantly less time reading our essays—is accidental, then we would have to run our study ten thousand more times to find it again. In other words, either we were struck by lightning in this particular experiment, or we’re on to something here. And just to make sure, Everett also ran what is called a “regression” analysis to determine if what we were detecting was partly being influenced by individuals’ level of political
knowledge—if this factor was involved shaping reading time. And it wasn’t. Rather, reading time was clearly related to conservatism, and especially its authoritarian and social conservative incarnation.
Given previous research, it may not be too outlandish to propose that this result may capture a general relative incuriosity that characterizes conservatives—although we cannot rule out the alternative hypothesis that they are just faster readers (a result that would surely generate no less of a stir!). Perhaps slightly more realistic is the hypothesis that conservatives are dismissive of the “liberal research enterprise,” and hence don’t deign to read our silly materials, but this is probably incorrect for two reasons. First, we threw out people who spent literally only a second or two on each page, so where participants truly didn’t care, their time-spent-reading measurements are excluded from this analysis. And second, being dismissive of the research enterprise is entirely consistent with being incurious anyway—heck, in a survey one might even measure incuriosity with an item that asks people whether they are dismissive of academic research.
So if reading time is a measure of curiosity, does it significantly correlate with Openness itself? Yes it does. It’s not a particularly strong relationship (r = 0.17, just significant at conventional levels: p = 0.04), but it’s suggestive that there is some relationship between the “curiosity” aspect of Openness and the general level of interest required to digest our essays.
We should emphasize that this finding about conservatism and reading time only held true in a group of 140 or so college students at a single university. But it is suggestive, especially since the result is found in all kinds of conservatives, is always statistically significant, and in some cases, is extremely so.
Furthermore, this result may signal a tendency in conservatives that cuts against our initial assumption that they engage in more or stronger motivated reasoning. As mentioned, spending more time reading essays or information that contradicts one’s point of view has been found, in prior studies, to be tied to more motivated reasoning. But conservatives in our study were spending less time reading across the board. This may have cut down on their sheer ability to be very biased, or motivated, in their responses. Many simply may not have been engaged enough with the material.
In sum, then, we simply didn’t achieve a good enough measure of general, non-political motivated reasoning to show that any one group of people is more likely to engage in it across a diversity of topics. This doesn’t mean such a measure could never be devised—only that our first attempt, into which we put much thought and even more work, didn’t succeed. To be sure, motivated reasoning happened all over the place in our survey, but it wasn’t systematic. The same individuals weren’t doing it on every item—rather, an individual tended to think in a biased way when he was heavily invested in that particular topic or in defending that particular attitude object, but not on other items in which he was less invested.
However, we did confirm the notion, resting now on a growing mountain of evidence, that liberalism is associated with curiosity and open-mindedness as measured by conventional methods—and if you consider time spent reading as an alternative measure, as measured by an alternative method too.
So what does this all mean?
While our hypothesis about conservatives engaging in more political motivated reasoning held up quite well in this study, a tendency to engage in more general motivated reasoning did not. However, we found one possible explanation for this result, in that conservatives, more than liberals, may have been going on quicker and less informed impressions rather than deeply engaging with the material we provided. It is even possible that conservatives were making more use of heuristics—which isn’t really reasoning at all, motivated or otherwise.
This suggestion itself arises out of past research on the differences between liberalism and conservatism. For instance, a study of authoritarianism and heuristic reasoning by Marcus Kemmelmeier, discussed in Chapter 3, suggested that this group of conservatives, in particular, was more susceptible to reasoning errors resulting from quick impulses or reactions to material. And again, our finding about less reading time most strongly implicated authoritarianism.
Also potentially relevant here is a study discussed in Chapter 8, on conservatives, global warming, and cable news. The study, by Lauren Feldman of American University and her colleagues, found that just as conservatives who watch Fox overwhelmingly dismiss global warming, so conservatives who watched CNN or MSNBC were more likely to accept that global warming is true. In other words, conservatives seemed more impressionable than liberals in both contexts.
In sum, our study very much backs up the idea that there may be something about conservatives that leads them to be more factually incorrect. But it also gives us a more nuanced view on the question, showing that we may not be able to locate this tendency simply in emotional defensiveness and the motivated reasoning that results. While highly sophisticated conservatives are likely very strong motivated reasoners about politics (and you can bet highly sophisticated liberals probably have this tendency too), average conservatives may be less exacting in how they assess information—less engaged, curious, exploratory—and more vulnerable to first impressions (including propaganda they encounter from trusted and intellectually sophisticated conservative opinion leaders). In other words, it may be mistaken to treat the two groups of conservatives in the same way in this context.
In future research, we would very much like to find new ways of testing these ideas, such as including measures of “need for cognition,” “need for closure,” and tests for various types of reasoning based on heuristics. It seems plausible that more and less reading time might be associated with the need for cognition and the need for closure, respectively.
As of now, we can still say that a lack of Openness probably explains much about many conservatives, including their resistance to the facts. But solely attributing this to an across the board difference in motivated reasoning that even extends outside of the political arena may be too simplistic—and thus, it is fortunate that we ran this study and were able to obtain this new evidence (and so modify our views). Instead, here is how Openness (or the lack thereof) might work:
If conservatives just aren’t as interested as liberals in finding things out about the world—and that’s what our essays were all about: we were purporting to bring evidence to bear about wide-ranging (and, we think, interesting) topics like ESP, quarterbacking ability, the academic quality of the participants’ school, a popular singer’s need to use performance enhancing technologies like auto-tune—one need not suggest conservatives are always more staunch defenders of ideas they care about than liberals are of ideas they care about. Conservatives’ tendency to be wrong on the facts might sometimes be explained by a lack of interest in facts themselves—and, perhaps, by a relatively stronger interest in seeing government set policy in a way that that matches their values (which are quite easily discoverable without any need for excessive curiosity), rather than changing on-the-ground realities.
In other words, even without a vast difference between liberals and conservatives in motivated reasoning, we can go a good distance in explaining why conservatives reject science and other evidence brought to bear on politics. It’s because Openness is largely a measure of curiosity about the world, an eagerness to inquire and learn new things—and that obviously often means inquiry about science and what’s verifiably true. It may be that in the course of reading a short essay attacking or praising some cherished idea, brand, or football team, liberals and conservatives alike can have equally strong, gut-level, emotional reactions, causing temporary denial. But over time, people who are curious about the world and more interested in learning about it are probably more likely to acquire knowledge—and, ultimately, to bring their political beliefs in line with that knowledge.
Such people are also, of course, probably more likely to wind up pursuing careers in “liberal” academia—and this book has presented much ev
idence showing that today’s Democratic Party, much more than today’s Republican Party, is brimming with intellectuals and Ph.D.s. So our study could also be said to reinforce a point made earlier in these pages—that the current, vast difference in expertise across the parties is probably closely related to their difference over what is true and false about the world. And this difference in expertise is itself surely related to Openness, and the tendency of the intellectually curious and exploratory to seek out knowledge and advanced degrees.
So it may be that greater Openness and greater interest in learning about the world in all its complexity—not a general lack of motivated reasoning—brings liberals closer to science and to the facts. And it may appear that conservatives are more motivated in their reasoning simply because, with policy preferences that are less likely to correspond to the kinds of knowledge that are acquired through curiosity and inquiry—and that thus are more likely to run afoul of evidence, or be oblivious to it—conservatives simply have a more frequent need to resort to political motivated reasoning to defend their beliefs.
Notes
241 “Everett Young” This chapter is co-authored by Everett Young and myself, as much of the text describing the study design and its findings was originally authored by him.
242 Ph.D. dissertation Everett Young, “Why We’re Liberal, Why We’re Conservative: A Cognitive Theory on the Origins of Ideological Thinking,” Ph.D. Dissertation, Stony Brook University, December 2009.
249 “1000 samples of similar size” This is almost like saying, regarding the correlation of .25 for Openness and ideology, that there was only a 1 in 500 (2 in 1,000) chance that we could have found that number by accident. But that is not quite right, so we can’t really tell you that.
It’s also similar to saying there’s a 499 in 500 chance that the correlation is “real” and that there’s a real relationship between liberalism and Openness, but that’s even further from being right. We’ll admit that, for simplicity, political scientists think of these “p-values” in these ways all the time, though we should try not to, for reasons that are beyond the scope of this book.
